Navigation system designs based on model predictive and sliding mode control for a driverless car's trajectory tracking

This study designs modern controllers, SMC and MPC, for vehicle motion control to ensure accurate trajectory tracking at the desired speed. Longitudinal motion is regulated via throttle and brake using a PD-FLC controller, while lateral motion is controlled through steering angle with SMC and MPC. The novelty of SMC lies in using Decoupled Algorithm, addressing a model where the steering input simultaneously affects translational dynamics and rotational dynamics, which is not widely applied in autonomous vehicles. The MPC novelty lies in using a precomputed adjustable coefficient matrix based on the receding horizon principle, simplifying design compared to continuous optimization during operation. Simulations in Matlab Simulink/Carsim under "Handling Course Left Edge" and "Double-Lane Change" showed that in Case 1, MPC achieved better performance metrics, with a 15% lower average distance error than SMC, while in Case 2, SMC outperformed MPC across all metrics, particularly in trajectory adherence, with 20% improvement.